STOCKS AND DEPRECIATION OF HUMAN CAPITAL: NEW EVIDENCE FROM A PRESENT‐VALUE PERSPECTIVE

Abstract

Responding to a perceived growing interest in human wealth estimates, this paper offers a framework for measuring the aggregate stock of human capital and then implements the procedure for the United States male population age 14 to 75. Unlike previous estimates of human wealth that are based upon historical or resource costs, these estimates measure the capital stock as the discounted resent‐value of expected lifetime returns. In the estimation, returns are equated with earnings data from the 1970 U.S. Census 15 percent Public Use Sample for out‐of‐school males, adjusted for employment and survival probabilities, adjusted for an assumed exogenous growth in future earnings, and discounted at 7.5 percent. We provide cross‐sectional estimates of individual stocks of human capital by age and educational attainment, as well as expected lifetime wealth profiles for individuals by level of education. These individual profiles can be used to obtain direct estimates of age‐specific depreciation which suggest human capital is subject to significant and prolonged appreciation before nearly straight‐line depreciation begins around middle age. This finding is all the more significant since resource‐cost estimates of human capital which must assume a depreciation pattern to obtain stocks have always imposed a much faster rate much sooner. Finally, an aggregate estimate of the stock of human capital for all males is supplied and its sensitivity to the choice of the discount rate, tax laws, and expected exogenous growth is analyzed. This seemingly‐conservative stock estimate is then compared to a much lower resource‐cost estimate offered recently by John Kendrick. A discount rate over 20 percent would be needed to equate the two measures. In trying to reconcile the two figures, we raise some new questions about the validity of both approaches for human capital accounting.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 69 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.